check_linux_bonding

Hi everyone,

Thank you for the nice Nagios plugin and the good work.
Here is my patch for version 1.3.2 after bonding driver changed in Linux 3.13.0.

Tested on Ubuntu (Linux 3.13.0-52-generic #86~precise1-Ubuntu SMP Tue May 5 18:08:21 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux):

I installed liblinux-kernelsort-perl to be able to compare kernel versions.

sudo apt-get install liblinux-kernelsort-perl

Then patch check_linux_bonding.


@@ -25,6 +25,7 @@ use strict;
use warnings;
use POSIX qw(isatty);
use Getopt::Long qw(:config no_ignore_case);
+use Linux::KernelSort;

# Global (package) variables used throughout the code
use vars qw( $NAME $VERSION $AUTHOR $CONTACT $E_OK $E_WARNING $E_CRITICAL
@@ -255,6 +256,15 @@ sub find_bonding_sysfs {
my $masters_file = "$sysdir/bonding_masters";
my @bonds = ();
my %bonding = ();
+ my $device;
+ my $kernel = new Linux::KernelSort;
+ my ($kernver, @rest) = split('-', `uname -r`);
+ my $retcmp = $kernel->compare($kernver,"3.13.0");
+ if ( $retcmp >= 0 ) {
+ $device = "lower";
+ } else {
+ $device = "slave";
+ }

if (! -f $masters_file) {
return {};
@@ -311,8 +321,8 @@ sub find_bonding_sysfs {

# get slave status
foreach my $slave (@slaves) {
- open my $STATE, '

0

It would appear the patch by BlauwBlaatje on April 29, 2014 has been truncated so is missing the final part of the patch. The following works for me:

--- check_linux_bonding.a 2014-10-13 11:13:17.899342000 +0100
+++ check_linux_bonding.b 2014-10-13 11:33:03.010804900 +0100
@@ -255,7 +255,14 @@
my $masters_file = "$sysdir/bonding_masters";
my @bonds = ();
my %bonding = ();
-
+ my $device;
+ my ($kernver, @rest) = split('-', `uname -r`);
+ if ( $kernver >= 3.13 ) {
+ $device = "lower";
+ } else {
+ $device = "slave";
+ }
+
if (! -f $masters_file) {
return {};
}
@@ -311,8 +318,8 @@

# get slave status
foreach my $slave (@slaves) {
- open my $STATE, '

As of Linux 3.13 the bonding driver uses /sys/class/net/bond0/lower_eth0/operstate instead of /sys/class/net/bond0/slave_eth0/operstate. Here is a patch (you might brush it up a bit ;))

--- check_linux_bonding.a 2014-04-30 09:22:10.523457651 +0200
+++ check_linux_bonding.b 2014-04-30 09:26:33.975182901 +0200
@@ -255,6 +255,13 @@
my $masters_file = "$sysdir/bonding_masters";
my @bonds = ();
my %bonding = ();
+ my $device;
+ my ($kernver, @rest) = split('-', `uname -r`);
+ if ( $kernver >= 3.13 ) {
+ $device = "lower";
+ } else {
+ $device = "slave";
+ }

if (! -f $masters_file) {
return {};
@@ -311,8 +318,8 @@

# get slave status
foreach my $slave (@slaves) {
- open my $STATE, '

Thank you for this excellent plugin !

You'll find below a patch of my own, allowing to ignore warnings if ad_num is different from number of registered slave.
In most cases, people would prefer being warned, but in my case, the same 802.3ad is bound on 2 different switches, generating 2x2 802.3ad, with one being "master", the 2 other links being "waiting" for a failure to become active.
This leads to being warned because 2 slaves over 4 seem to be missing in the active 802.3ad bonding, which is half true and half false.
I therefore don't wan't to be warned.

Regards,
Benjamin


--- check_linux_bonding.orig 2012-07-24 10:52:55.973316334 +0200
+++ check_linux_bonding 2012-07-24 11:10:44.681319464 +0200
@@ -78,6 +78,7 @@
-n, --no-bonding Alert level if no bonding interfaces found [ok]
--slave-down Alert level if a slave is down [warning]
--disable-sysfs Don't use sysfs (default), use procfs
+ --ignore-num-ad Don't warn if num_ad_ports != num_slaves
-b, --blacklist Blacklist failed interfaces
-d, --debug Debug output, reports everything
-h, --help Display this help text
@@ -110,6 +111,7 @@
'linebreak' => undef,
'verbose' => 0,
'disable_sysfs' => 0,
+ 'ignore_num_ad' => 0,
'slave_down' => 'warning',
);

@@ -124,6 +126,7 @@
'linebreak=s' => $opt{linebreak},
'v|verbose' => $opt{verbose},
'disable-sysfs' => $opt{disable_sysfs},
+ 'ignore-num-ad' => $opt{ignore_num_ad},
'slave-down=s' => $opt{slave_down},
) or do { print $USAGE; exit $E_UNKNOWN };

@@ -490,7 +493,7 @@
$b, $bonding{$b}{mode};
report($msg, $E_CRITICAL);
}
- elsif (defined $bonding{$b}{ad_num} and $bonding{$b}{ad_num} != scalar keys %slave) {
+ elsif ($opt{ignore_num_ad} == 0 and defined $bonding{$b}{ad_num} and $bonding{$b}{ad_num} != scalar keys %slave) {
my $msg = sprintf 'Bonding interface %s [%s]: Number of AD ports (%d) does not equal the number of slaves (%d)',
$b, $bonding{$b}{mode}, $bonding{$b}{ad_num}, scalar keys %slave;
report($msg, $E_WARNING);

This is a great piece of work. Thank you for making it available.